Add reverse fine-tuning model for structured data extraction #1
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This commit adds a reverse model that extracts structured metadata from natural language payment descriptions, opposite to the original model which generates natural language from structured data. New files: - generate_reverse_dataset.py: Creates reverse training dataset - finetune_phi3_reverse.py: Fine-tuning script for reverse model - test_reverse_model.py: Testing and inference script - README_REVERSE.md: Comprehensive documentation - reverse_payments_dataset/: Generated training data (500 examples) The reverse model enables: - Parsing user payment descriptions into structured formats - Building conversational payment interfaces - Extracting payment data from emails/notifications - Converting natural language to API-ready formats Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
Updates: - Enhanced README.md with dual-model information - Added ABOUT.md with repository description and stats - Documented branch structure and model differences - Added repository topics for discoverability Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
Updates to README_REVERSE.md: - Complete restructure with clear Quick Start guide - Added real-world examples and use cases - Detailed integration code with FastAPI example - Parsing utilities for structured output - Troubleshooting specific to reverse model - Dataset customization examples - Performance benchmarks and cross-references - Clear file naming (generate_reverse_dataset.py, etc.) - Branch navigation instructions Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This commit adds a reverse model that extracts structured metadata
from natural language payment descriptions, opposite to the original
model which generates natural language from structured data.
New files:
The reverse model enables:
Generated with Claude Code
Co-Authored-By: Claude noreply@anthropic.com